Research map / Vision-language and multimodal / Human and embodied modeling
ConvNTM: Conversational Neural Topic Model
Hongda Sun, Quan Tu, Jinpeng Li, Rui Yan
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add conversation sequence encoder for multi-turn context (ablation) | ConvNTM · DailyDialog | TQ: 0.4462 -> 0.4907 | Improved |
| add multi-role graph encoder (ablation) | ConvNTM · DailyDialog | TQ: 0.4429 -> 0.4907 | Improved |
| model multiple speaker roles instead of treating all utterances as one speaker (ablation) | ConvNTM · DailyDialog | TQ: 0.3964 -> 0.4907 | Improved |
| add word co-occurrence objective (ablation) | ConvNTM · DailyDialog | CV: 0.6237 -> 0.6542 | Improved |
| add word co-occurrence objective (ablation) | ConvNTM · DailyDialog | TD: 0.780 -> 0.750 | Hurt |
| replace GNTM topic labels with ConvNTM topic labels | STM dialogue act classifier · DailyDialog dialogue act classification | Accuracy: 87.2 -> 88.9 | Improved |
| use ConvNTM topic words instead of GNTM topic words | THERD response generator · DailyDialog response generation | PPL: 36.68 -> 34.14 | Improved |
| use ConvNTM topic words instead of GNTM topic words | THERD response generator · DailyDialog response generation | BLEU-1: 18.53 -> 20.14 | Improved |
| use ConvNTM topic words instead of GNTM topic words | THERD response generator · DailyDialog response generation | Distinct-1: 8.26 -> 8.79 | Improved |
Closest papers by meaning
- A Knowledge-Grounded Neural Conversation Model (AAAI 2018 · similarity 0.87)
- Neural Dynamic Focused Topic Model (AAAI 2023 · similarity 0.87)
- NaturalConv: A Chinese Dialogue Dataset Towards Multi-turn Topic-driven Conversation (AAAI 2021 · similarity 0.87)
- textTOvec: DEEP CONTEXTUALIZED NEURAL AUTOREGRESSIVE TOPIC MODELS OF LANGUAGE WITH DISTRIBUTED COMPOSITIONAL PRIOR (ICLR 2019 · similarity 0.86)
- Topic-Aware Multi-turn Dialogue Modeling (AAAI 2021 · similarity 0.86)
- Tuning LayerNorm in Attention: Towards Efficient Multi-Modal LLM Finetuning (ICLR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
